Ketone Incorporation Extends the Emission Properties of the Xanthene Scaffold Beyond 1000 nm

Abstract

Imaging in the shortwave-infrared region (SWIR, $\lambda =1000-2500\mathbf{n}\mathbf{m}$) has the potential to enable deep tissue imaging with high resolution. Critical to the development of these methods is the identification of low molecular weight, biologically compatible fluorescent probes that emit beyond 1000 nm. Exchanging the bridging oxygen atom on the xanthene scaffold (C10’ position) with electron withdrawing groups has been shown to lead to significant redshifts in absorbance and emission. Guided by quantum chemistry computational modeling studies, we investigated the installation of a ketone bridge at the C10’ position. This simple modification extends the absorbance maxima to 860 nm and the emission beyond 1000 nm, albeit with reduced photon output. Overall, these studies demonstrate that broadly applied xanthene dyes can be extended into the SWIR range.

INTRODUCTION

Fluorescence-based methods have transformed modern biological research. Shifting absorbance and emission to longer wavelengths reduces competitive excitation of biomolecules and light scattering - both of which can improve imaging quality. This has led to a concerted effort to identify long-wavelength emitting fluorophores for use in experiments ranging from cellular analysis to in vivo imaging (1,2). Recent efforts have shown that the shortwave infrared (SWIR, $\mathrm{\lambda }=1000-2500\mathrm{n}\mathrm{m}$) or near-infrared II (NIR-II) region provides enhanced resolution and depth penetration in bulk tissue (3,4). Critical to the development of these methods was the availability of InGaAs detector-based cameras. These hardware developments have been complemented by significant progress in the design and synthesis of dyes that absorb and emit above 1000 nm (5–7). However, existing probes involve challenging multistep synthesis and exhibit low fluorescence quantum yields (8,9). Overall, there remains a significant need for novel, biologically compatible, small-molecule fluorophores in this range.

Rhodamine and fluorescein dyes are important fluorophores and foundational components of many cellular imaging experiments. This is due to their excellent cell permeability, low toxicity and fluorogenic properties allowing for imaging cellular processes in real time (10). One strategy for extending the absorbance and emission maxima of these dyes involves extending the π-system, which can lead to an undesirable increase in hydrophobicity and molecular weight (11). Another more subtle strategy is to alter the C10’ bridging atom of the xanthene scaffold (Fig. 1A). Prior reports have shown that modifying the bridging oxygen atom on the xanthene scaffold at the C10’ position can lead to significant redshifts in absorbance and emission (12–23). In particular, recently reported phospho- and sulforhodamine derivatives have been reported with emission up to ~740 nm (Fig. 1A).

Figure 1.

(A) Computational and experimental absorbance/emission maxima data for substituents (1–7) at R-position on a dimethylamino-xanthene-derived scaffold (see Table S1 for X-group substituent). (B) Comparisons between the HOMOs and LUMOs of R1-R7. (C) Correlation between experimental and predicted ${\mathrm{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ absorbance using TDDFT-B3LYP-PCM(H₂O) and ORMAS-PT2-PCM(H₂O). (D) Correlation between the computed ${\mathrm{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ absorbance (TDDFT-B3LYP-PCM(H₂O)) and HOMO/LUMO orbital energies. (E) Correlation ($R^2=0.94$) between experimental ${\mathrm{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ absorbance and C–C10’ bond order in the LUMO. (F) Correlation ($R^2=0.97$) between signed R-group Mulliken population in LUMO and computed ${\lambda }_{\mathrm{m}\mathrm{a}\mathrm{x}}$ absorbance. (G) Comparison between computed absorbance and emission oscillator strengths (TDDFT-B3LYP-PCM(H₂O)) for R1-R7. H) Comparison between computed absorbance oscillator strengths (TDDFT-B3LYP-PCM(H₂O)) for R1-R7 and experimental extinction coefficient values for structurally related compounds (see Supporting Information for further details).

Here, we report the computational design, synthesis and initial analysis of xanthene derivatives substituted with a ketone at the C10’ position. Across a series of known C10’-substituted systems, quantum chemical predictions of absorbance and emission maxima are in good agreement with experiment, and further calculations indicate that a ketone moiety would lead to a dramatic redshift. Additional analyses of the molecular orbitals involved in the electron excitations clearly demonstrate the link between substituent πelectronegativity and bathochromic shift. While a C10’ ketone-substituted rhodamine had been described in a single report in the patent literature (24), no detailed information was available. Here, we report optimized syntheses of ketone-substituted xanthenes and rhodamines, as well as amide derivatives. Additionally, we detail that di-halogenated derivatives can undergo a bis-Heck reaction to provide pH-responsive exo-olefin derivatives. These synthetic efforts allowed key photophysical properties of these probes to be determined. Overall, these studies reveal the significant potential, as well as limitations, of these redshifted xanthene-based dyes as imaging agents in the SWIR range.

MATERIALS AND METHODS

Computational chemistry.

Quantum chemistry calculations were performed using a local version of the GAMESS package (25,26), where cc-pVTZ basis sets (spherical harmonics) were used throughout (27,28). Molecular orbitals were illustrated with MacMolPlt (29). Density functional theory (DFT) (30) and time-dependent density functional theory (TDDFT) (31–33) utilizing the B3LYP (34–36) functional were used to compute absorbance (S0 → S1*) and emission (S1 → S0*) (37) energies. The occupation restricted multiple active space (ORMAS) (38,39) method with second-order perturbation theory correction (ORMAS-PT2) (40) was also used to compute absorbance and emission energies at B3LYP and TDDFT-B3LYP optimized geometries respectively. ORMAS active spaces included all valence π-electrons and orbitals and state-averaged energies were optimized. Water-solvent effects were included in all calculations via the polarizable continuum model (PCM) (41–45) approach. Mulliken electron population analysis (46) was used to analyze the highest occupied molecular orbitals (HOMOs) and lowest unoccupied molecular orbitals (LUMOs). Full details and results, including ORMAS active space partitioning scheme, are available in the Supporting Information.

Synthetic chemistry.

Detailed experimental procedures and characterization data are available in the Supporting Information.

Determination of molar absorption coefficients and fluorescence quantum yield of KR-1.

All spectroscopic measurements were carried out in duplicate at room temperature using 10 mm × 10 mm quartz cuvettes (Hellma GmbH) and high-purity spectroscopic grade solvents. High-purity water $(\mathrm{\sigma }<55\mathrm{\mu }\mathrm{S})$ was obtained using a MilliPore filter system. Methanol (MeOH; Uvasol), Dimethylsulfoxide (DMSO; spectroscopic grade) and Dichloromethane (DCM; anhydrous) were purchased from Merck. The reference dye IR140 that was previously studied by us (47) was purchased from Lambda Physik.

Absorption measurements for the quantum yield determination were carried out on a calibrated absorption spectrometer Cary 5000 from Varian, Inc., using a step width of 1 nm, an integration time of 0.1 s and a spectral bandwidth of 2 nm. The absorbance spectra are corrected for blank absorption and scattering. The absorbance of the dye solutions used for fluorescence quantum yield determinations was generally kept below 0.1 to minimize dye aggregation and reabsorption effects for the measurements of the emission spectra and integral emission intensities (48).

Fluorescence emission measurements were performed with a calibrated spectrofluorometer FLS-920 from Edinburgh Photonics, equipped with a Xenon lamp, double monochromators, and a nitrogen-cooled PMT R5509P from Hamamatsu. A 550 nm cut-on filter was used in the excitation light path. The emission spectra were obtained by averaging 2 to 4 scans of the emission spectra (depending on the signal-to-noise ratio) with a step width of 1 nm, an integration time of 2 s and excitation and emission slit widths of 16 nm and 8 nm. Magic-angle conditions with the excitation and emission polarizers set to 0° and 54.7°, respectively, were employed to render the measured emission intensities independent of the emission anisotropy of the NIR emitter. All emission spectra were corrected for excitation intensity fluctuations using the signal of a reference detector as well as for blank emission and scattering by subtraction of the solvent spectra and for the wavelength dependence of the instrument’s spectral responsivity utilizing a previously determined emission correction curve (48). The instrument was flushed with nitrogen during the measurements to avoid light absorption from atmospheric water (47).

The fluorescence quantum yield was determined relatively to the dye IR140 following a published procedure (48), the fluorescence quantum yield of which had been previously determined absolutely and relatively with different setups (49), and calculated according to the following equation:

$${\mathrm{\varphi }}_x={\mathrm{\varphi }}_{ref}\frac{F_x{f}_{ref}\left({\mathrm{\lambda }}_{ex,ref}\right)n_x^2}{F_{ref}{f}_x\left({\mathrm{\lambda }}_{ex,x}\right)n_{ref}^2}$$

where $\varphi$ is the fluorescence quantum yield, $F$ is the integral photon flux, $f$ is the absorption factor at the excitation wavelength ${\lambda }_{ex}$ and $n$ is the refractive index of the solvent. The indices $x$ and $ref$ denote the sample and reference respectively. For the reference dye IR140, a fluorescence quantum yield of 0.20 in DMSO was used that was previously measured absolutely with a calibrated and validated integrating sphere setup (49). The absorption factor $f$ at the excitation wavelength was obtained by averaging the absorption over the excitation bandwidth. The spectral width of the spectrofluorometer’s excitation monochromator was not considered. An excitation wavelength of ${\lambda }_{\mathrm{e}\mathrm{x}}=780\mathrm{n}\mathrm{m}$ was used for the dyes IR140 and KR-1. The refractive indices that were employed in the calculations were 1.479 for DMSO and 1.424 for DCM. The integrated photon fluxes required for the determination of the number of emitted photons were obtained in the wavelength ranges 970–1200 nm for KR-1 and IR140.

For additional details, including extinction coefficients, relative quantum yield measurements with KR-2 and KR-3 and details of the measurements taken in Fig. 2, see Section S5.

Figure 2.

(A) Normalized absorption (dashed) and emission (solid) data of KR-1 and KR-3 in CH₂Cl₂. (B) Normalized absorption (dashed) and emission (solid) data of KR-1 in PBS pH 7.4. (C) Normalized absorption (dashed) and emission (solid) data of KR-3 in PBS pH 7.4. (D) Normalized absorption (dashed) and emission (solid) data of KR-2 in CH₂Cl₂ (black trace) and CH₂Cl₂ supplemented with 5% TFA (red trace). (E) Exo-KR-4 (250 μM) in PBS buffered from pH 7.5 (black trace) to pH 3.5 (red trace) to generate KR-4. (F) Plot of absorbance intensity at ${\lambda }_{\mathrm{m}\mathrm{a}\mathrm{x}}=875\mathrm{n}\mathrm{m}$ vs pH; sigmoidal plot fit resulted in apparent ${\mathrm{p}\mathrm{K}}_{\mathrm{a}}=4.4$.

RESULTS AND DISCUSSION

Computational design

We first sought to identify the role of the central R-group fused at the C10’ position in dimethylamino-xanthene-derived scaffolds (Fig. 1A). While our main goal was to identify a chemical modification that would significantly redshift absorbance and emission, we also sought to correlate spectroscopic properties with R-group electronic effects. We first computed absorbance and emission energies of the known series of systems R1-R6 and the ketone derivative R7 (Fig. 1C, Figure S4 and Table S4). We found the ORMAS-PT2-PCM(H₂O) method to be very reliable (absorbance/emission mean unsigned errors, MUEs, of 0.065/0.039 eV) and the TDDFT-B3LYP-PCM(H₂O) theory slightly less so (absorbance/emission MUEs of 0.397/0.167 eV). Importantly, both methods ranked the fluorophores in the correct order and, perhaps surprisingly, computed emission energies were generally more accurate than absorbance data. Most notably, the ketone derivative R7 was predicted to absorb/emit above 800/900 nm—a dramatic bathochromic shift relative to other systems. As described below, this prediction was subsequently confirmed experimentally.

After verifying that all singlet states had predominant HOMO (π) → LUMO(π*) characters (see SI for TDDFT-B3LYP-PCM excitation/deexcitation amplitudes), we further analyzed these orbitals. While HOMO structures and energies remain consistent across the series R1-R7 (Fig. 1B,D), LUMO energies decrease substantially (by ~25 kcal mol⁻¹, Fig. 1D). The gradual change in LUMO character is visually apparent where C–C10’ antibonding nodes are easily discernable for R1 and R2 but not for the other substituents that show a gradual extension of the bonding lobe to the C10’ position for R3 → R6 and beyond for R7 (Fig. 1B). Indeed, quantified C–C10’ bond orders in the LUMOs (Table S5) correlate very well with absorbance energies (Fig. 1E), thus rationalizing LUMO stabilization across the series. Furthermore, signed R-group LUMO electron populations (50) increase steadily across the series (Table S6) and show an excellent correlation with the absorbance energies (Fig. 1F), indicating that excitation energies are lowered as the R-group π-electronegativity increases. In complete contrast, HOMO C-C10’ bond orders and R-group populations are essentially zero throughout (Table S6). These observations can be reconciled by noting that: (1) HN− and O− substitutions add two electrons to the π-system, thus destabilizing the LUMO, (2) the other R-groups except OC—stabilize the LUMO by withdrawing some π-electron density and (3) OC—substitution adds two electrons and an empty antibonding π-orbital that the LUMO can extend into.

Computed TDDFT-B3LYP-PCM(H₂O) oscillator strengths rise steadily across the series R1 → R4, but then suddenly decrease with the ketone derivative R7 predicted to have the lowest absorption and emission intensities (Fig. 1G). Notably, this trend is confirmed experimentally with a strong correlation between oscillator strengths and absorption coefficients (Fig. 1 H), which includes the ketone-substituted compound detailed below. Further studies are needed to determine the origin of the trend between oscillator strengths and the C10’ functional group—particularly for the design of brighter systems. While the computational chemistry results for the substituted series R1-R7 are themselves illuminating, these studies clearly suggest that ketone substitution should yield a fluorophore with dramatically bathochromic-shifted absorption and emission bands.

Synthesis of keto-rhodamine (KR) dyes

Initial synthetic efforts focused on generating the parent ortho-methyl derivative KR-1 using previously reported conditions (Scheme 1A) (24). We found that we could prepare KR-1 in yields between 5% and 8% through the AlCl₃-promoted addition of 1 to 2-methylbenzoyl chloride. A screen of Lewis acids did not identify improved conditions. Although this reaction provided KR-1, it was difficult to purify and low yielding. Consequently, we considered alternative routes and were particularly inspired by a recent report by Lavis and coworkers that detailed the organolithium-mediated route to rhodamine derivatives (51). Building on this report, we developed the synthetic scheme shown in Scheme 1B, which could be used to generate KR-1 in 25% yield over six steps. The synthesis started with methylation of bis(3-aminophenyl)methanone 6 with iodomethane in the presence of Cs₂CO₃ to give 1. This was followed by regioselective bis-bromination alpha to the ketone bridge to generate dihalogenated 7. The ketone 7 was reduced using NaBH₄ before subsequent alkylation to give a MOM (2a), Me (2b) and TBS (2c) protected intermediates 2a-c respectively. These intermediates were screened for their ability to undergo lithiation and bisaddition to methyl 2-methylbenzoate. Both the MOM (2a) and Me (2b) protected compounds provided 3a and 3b in good yields of 58% and 54% respectively. Interestingly, the t-butyldimethylsilyl protected 2c underwent transmetalation followed by silyl protecting group migration in a Retro-Brook-type rearrangement to give a organosilicon (C-Si) product (see Supporting Information for further details) (52). Upon deprotection using either TFA or BBr₃, KR-1 could be accessed in 25% overall yield from the MOM protected 3a or 22% for the Me 3b, a ~ 3-fold improvement on the previously reported route.

Scheme 1.

(A) Friedel–Crafts approach to KR-1. ^a(1 eq.), 2-methylbenzoyl chloride (1.1 eq.), AlCl₃ (1.75 eq.), solvent (2–2.5 M). ^b1 (1 eq.), 2-methylbenzoyl chloride (1.1 eq.), Lewis acid (1.75 eq.) and CH₂Cl₂ (2 M). (B) Organolithium-mediated approach to KR-2 and KR-3. (C) Double Heck Approach to KR-4. See Sections S3 and S4 for detailed synthetic procedures and analysis.

Given the benefits of the late stage deprotection, the MOM-protected compound 2a was chosen as our key intermediate. Phthalic anhydride underwent bis-aryl addition to generate 4 which remained as a ring-opened 9,10-anthracenediyl derivative until subsequent deprotection with TFA to give ring-closed KR-2 in 27% overall yield (Scheme 1B). We also set out to prepare a water-soluble “always on” amide derivative KR dye, as related probes have been described extensively in the rhodamine literature (53–60). The carboxylic acid 4 underwent amide coupling under standard conditions to give doubly protected intermediate 9. Upon treatment with DDQ, 9 underwent MOM deprotection and in situ oxidation before standard tert-butyl ester deprotection with TFA in CH₂Cl₂ gave title compound KR-3 in 18% yield over eight steps (Scheme 1). While TBS-protected dihalogenated 2c was not suitable for Li-Br exchange and subsequent bis-addition, we hypothesized that it might undergo a palladium-mediated tandem Heck reaction with a protected acrylate species. Prior reports of similar chemistry have been applied to the generation of xanthene and anthracene derivatives from o, o’-dibromobiaryls with ethyl acrylate (61). However, to our knowledge this strategy has not been applied to generate rhodamine-based derivatives. Initially, methyl acrylate was chosen to screen reaction conditions and it was found employing Pd (PPh₃)₄, TEA and DMF at 150°C provided TBS-protected 10 in 40% yield (see Supporting Information). This product likely results from intermolecular Heck followed by an intramolecular Heck reaction to form the xanthene core. Deprotection was achieved using the mild conditions of ACN: H₂O: formic acid (2:1:0.05) to give exocyclic 11. Upon subjecting 11 to an acidic environment (1:1 MeCN:H₂O w/1% formic acid), a longwavelength (${\mathrm{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}=870\mathrm{n}\mathrm{m}$) absorption band appeared (Figure S1). Encouraged by this result and in an effort to synthesize water-soluble conjugatable derivatives, compound 2c was subjected to the same tandem-Heck ring-closing conditions with tert-butyl acrylate affording 5 in 71% yield (Scheme 1C). 5 underwent tert-butyl ester deprotection in TFA in CH₂Cl₂ and subsequent TSTU mediated activated ester formation before coupling with a heterobifunctional pegylated linker. Finally, using the same mild deprotection conditions of ACN: H₂O: formic acid (2:1:0.05), exocyclic KR-4 could then be accessed in 63% yield from 5.

Photophysical properties

Initial photophysical measurements concentrated on comparing the always-ON KR-1 and KR-3 to the parent ether-bridged reference dye tetramethylrosamine (Ros) (62). KR-1 exhibited absorption $\left({\mathrm{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}\right)$ and emission $\left({\mathrm{\lambda }}_{\mathrm{e}\mathrm{m}}\right)$ maxima of 862/1058 nm respectively (Table 1). This equates to bathochromic shifts in absorbance/emission of 313/488 nm relative to Ros in PBS. KR-1 and KR-3 showed similar ${\mathrm{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ and ${\mathrm{\lambda }}_{\mathrm{e}\mathrm{m}}$ values in CH₂Cl₂ and a similarly pronounced redshift in ${\mathrm{\lambda }}_{\mathrm{e}\mathrm{m}}$ of 1065 nm was observed for KR-3 in PBS 7.4 (Table 1 and Fig. 2A–C). Notably, KR-1 and KR-3 exhibit significantly reduced absorption coefficients, which is in line with the computationally derived oscillator strengths (see above). The fluorescence quantum yields were determined to be 0.013 and 0.0031 in CH₂Cl₂, respectively, and were approximately two orders of magnitude lower in PBS. Therefore, this dramatic bathochromic shift in absorbance and emission is accompanied by significantly reduced photon output. This is to be expected according to the energy gap rule, which states that a smaller energy gap between S₁ and S₀ favors internal conversion relative to fluorescence (63). Thus, the reduced quantum yield is due to solvent-mediated nonradiative deactivation pathways—a feature also found in other SWIR dyes (4,8). Notably, however, previously reported cyanine-based SWIR probes maintain excellent absorption coefficients (64–66). This leads to overall brighter dyes, with brightness defined as the product of the absorption coefficient at the excitation wavelength and quantum yield—a useful metric to determine the magnitude of the observable fluorescence signal.

Table 1.

Optical properties measured in dichloromethane (CH₂Cl₂) and in pH 7.4 PBS.

Entry	${\lambda }_{\mathrm{abs}}$ (nm) (CH2Cl2 / PBS)	${\lambda }_{\mathrm{em}}$ (nm) (CH2Cl2 / PBS)	$\mathrm{\epsilon }$ (M−1 cm−1) (CH2Cl2 / PBS)	${\mathrm{\Phi }}_{\mathrm{F}}$ (CH2Cl2 / PBS)	$\mathrm{\epsilon }\times {\mathrm{\Phi }}_{\mathrm{F}}$ (PBS)	${\tau }_{\mathrm{i}}$ (ns) (CH2Cl2)
Ros*	–/549	–/570	–/72 000	–/0.31	22 320	-
KR-1	855/862	911/1058	28 400/16 500	0.013/0.00013	2.15	0.346
KR-2	848†/–	907†/–	4700/–	0.00029/–	1.36	-
KR-3	855/870	919/1065	15 600/9500	0.0031/0.00017	1.62	-
KR-4	–/874‡	–/1058‡	–/1600‡	-	-	-

^*Ref. (62)

^†CH₂Cl₂ w/ 5% TFA

^‡acetate buffer (0.05 M) pH 3.6. The quantum yield (Q.Y.) of KR-1 was measured against IR-140 in DMSO as a reference (assuming a Q.Y. of 20%). KR-1 was used as a standard for subsequent Q.Y. measurements. See Section S5 for additional data and discussion.

In addition to conventional photon output, a useful feature of rhodamine dyes is their fluorogenic properties. Specifically, an equilibrium exists between a spirolactone “ring-closed,” nonfluorescent form and “ring-opened” fluorescent species (67,68). Having synthesized KR-2 and KR-4, their fluorogenic switching properties were examined (Fig. 2D,G). In 5% TFA CH₂Cl₂, KR-2 underwent ring opening as indicated by the increase in ${\mathrm{\lambda }}_{\mathrm{m}\mathrm{a}\mathrm{x}}$ 848 nm and ${\mathrm{\lambda }}_{\mathrm{e}\mathrm{m}}$ at 907 nm respectively. Notably, TFA was required to induce the appearance of the NIR absorption band and we could not identify other conditions that led to the ring-opened form.

An additional feature of certain xanthene derivatives is an exocyclic double bond (69,70). This was recently investigated by Johnsson and coworkers, where photoactivatable silicon rhodamines possessing an exocyclic alkene were generated through bisaryl lithium addition to glutaric anhydride (69). Our palladiummediated approach to KR-4 provided a simple way to prepare ester-substituted exoalkenes. NMR studies of KR-4 in 9:1 CD₃CN, D₂O revealed a proton singlet at 6.60 ppm which showed strong HSQC coupling to a CH unit at 119.5 ppm (see Supporting Information). Additionally, KR-4 exhibits a ${\mathrm{\lambda }}_{\mathrm{a}\mathrm{b}\mathrm{s}}$ of 455 nm in PBS at pH 7. These results suggest that under neutral conditions, these compounds prefer the exoform. We hypothesized that acidic conditions might promote the conversion to an endoform, which would reestablish the long-wavelength optical properties of the xanthene core. In practice, we found that decreased pH led to the formation of new absorbance maxima ${\mathrm{\lambda }}_{\mathrm{a}\mathrm{b}\mathrm{s}}$ of 874 nm with a ${\mathrm{p}\mathrm{K}}_{\mathrm{a}}$ of 4.4 (Fig. 2E–G). These studies suggest this endo/exoisomerization chemistry can be used as a bioresponsive fluorogenic mechanism in the physiological pH range—an observation which may extend to other xanthene-based probes.

CONCLUSION

Here, we describe the longest emitting xanthene derivatives reported to date. This was achieved by the replacement of the ether bridge with a π-electron withdrawing ketone functional group on the xanthene core, leading to dramatic bathochromic shifts in absorbance and emission maxima. These studies demonstrate the significant impact of π-electron-withdrawing groups at this position on the electronic structure of the xanthene core. While carboxylic acid substituted KR-2 remained in the nonfluorescent lactone form under biological conditions, their methyl and tertiary amine counterparts KR-1 and KR-3 exhibited large Stokes shifts and absorption and emission maxima that extend into the SWIR region at physiological pH. A fluorogenic exoolefin KR-4 was synthesized and was shown to undergo an acid mediated exo–endo switching mechanism.

Broadly, these studies reveal that ketone substitution can have a dramatic effect on the absorption and emission maxima of xanthene dyes like rhodamines. However, the moderate absorption coefficients and quantum yields of these molecules need to be considered in the design of future strategies that seek to apply these probes. In this regard, we anticipate that the low molecular weight of these molecules suggests that multimerization approaches have significant potential.

Supplementary Material

34676539_Schnermann_SM1

Section S1. Supporting figures and tables.

Table S1. Structures, photophysical properties and literature references of previously reported compounds in Fig. 1A.

Figure S1. Exo- to endo-switching of 11 (200 μM) in acidic media.

Figure S2. Stacked ¹H NMR spectra of compounds 2c, 2c-1, 2c-2 and 2c-3 in ACN-d3.

Section S2. General materials and methods.

Section S3. Synthetic procedures.

Section S4. Retro-Brook rearrangement experiment and NMR analysis.

Scheme S1. Synthetic scheme of H2O/ D2O quench experiment with predicted products 2c-1 and 2c-2.

Table S2. Structures, photophysical properties and literature references of dyes used in absorbance oscillator strengths (S0 → S1*) and (S1*→ S0) vs extinction coefficient (Fig. 1E/F).

Figure S3. Stacked ¹H NMR spectra of compounds 2c, 2c-1, 2c-2 and 2c-3 in ACN-d3 (zoom 7.5–3.3 ppm).

Section S5. Photophysical measurements.

Section S6. ¹H, ¹³C and associated 2D NMR spectra.

Section S7. Computational chemistry.

Table S3. Numbers of orbitals in each group and associated occupancy minima

Table S4. Computed absorbance and emission energies (in eV with oscillator strengths in parentheses) vs experimentally measured. Calculations included water-solvent effects via the Polarizable Continuum Model (PCM). Mean unsigned errors (MUEs) for each method shown at bottom. See Methods for full details.

Figure S4. Experimental vs predicted absorbance (A) and emission (B) energies (eV) using TDDFT-B3LYP-PCM(H₂O) and ORMAS-PT2-PCM(H₂O) methods. Mean unsigned errors (MUEs) in absorbance/emission predictions are 0.397/0.167 and 0.065/0.039 for TDDFT-B3LYP and ORMAS-PT2 respectively.

Table S5. Experimental and computed (TDDFT-B3LYP-PCM) absorbance energies (eV), C–C10’ bond orders and signed R-group electron populations in the LUMOs. Properties determined at ground-state optimized geometries. Calculations included water-solvent effects via the polarizable continuum model (PCM). See Methods for full details. Correlation coefficients $\left(R^2\right)$ between experimental (left) and computed (right) absorbances vs LUMO C–C10’ bond orders and signed R-group electron populations are shown at bottom.

Table S6. Examinations of HOMOs and LUMOs in the C10’ substituted series via Mulliken electronic population analysis. For each orbital, the C–C10’ bond order is provided along with the total electron population on the R-group. Analyses were performed at ground-state optimized geometries.

Section S8. Raw computational chemistry data.

Section S9. References.